Open-source software compares Abel transforms for easy use
Open-source software compares Abel transforms for easy use lead image
Abel transforms are mathematical functions used to convert 2-D data into the three dimensions inherent in the object they represent. But choosing the best Abel transform to use is tedious, time-consuming and involves trial and error. Comparing the available methods and choosing the best one could take years, and “a lot of data gets thrown out, because the pre- and post-processing tools aren’t up to the task” said study author Daniel Hickstein.
Hickstein et al.—an international, interdisciplinary team of collaborators who met online and discovered they had similar needs—tested a series of Abel transforms and characterized each method, noting how well they preserve resolution and deal with noise in image data. They translated the eight methods into Python, combined them in a software package called PyAbel, and added tools like centering algorithms, a circularization algorithm and an angular integration function, among others.
“A lot of the existing algorithms couldn’t handle large images. You would basically just run out of memory, or it would take forever for images more than a few megapixels,” Hickstein said.
In the process of rewriting and optimizing the Abel transforms, Hickstein’s team sped up their performance and unified them to use the same data formats and mathematical conventions.
“We had to find out what the rate limiting step was and how we could rephrase that math to call functions from efficient math libraries,” Hickstein said. “The speed improvement makes it possible to apply Abel transforms to huge datasets consisting of thousands or millions of images.”
PyAbel is open-source software available on GitHub, along with detailed documentation.
The authors hope it will ease data-processing workflows in fields ranging from photoelectron spectroscopy to analyzing flames and plasma plumes from lasers, as well as large astronomy datasets.
Source: “A direct comparison of high-speed methods for the numerical Abel transform,” by Daniel D. Hickstein, Stephen T. Gibson, Roman Yurchak, Dhrubajyoti D. Das, and Mikhail Ryazanov, Review of Scientific Instruments (2019). The article can be accessed at http://doi.org/10.1063/1.5092635 .
